Re: [Boost] [Describe] Summary Review Summary

From: Robert Ramey (ramey_at_[hidden])
Date: 2021-03-16 18:55:05

On 3/16/21 9:49 AM, Peter Dimov via Boost wrote:
> Niall Douglas wrote:
>> Most libraries presented for standardisation ARE battle hardened libraries.
>> However they were typically written for preceding C++ standards, and in
>> outdated idioms and design patterns, and require modernisation which can
>> involve substantial refactoring.
> 
> Maybe not most.
> 
> It's no longer easy to give a single reason as to why the committee's
> output is what it is, as the submissions are of varying quality, philosophy, and
> origin; there's no single universal problem with them.
> 
> Some of them come from big companies, are indeed totally battle-hardened,
> but reflect the monoculture of the company which does not correspond
> very well to the environment in which the median C++ programmer operates.
> 
> Others are useful facilities in use in more than one relatively smaller company,
> battle-hardened, but pragmatic, prioritizing problem-solving over design
> elegance or avoidance of undefined behavior.
> 
> And then we still have the "direct to video" libraries that haven't actually
> seen much practical use.
> 
> In theory the committee should be that great place where people who value
> design elegance and lack of undefined behavior and specification imprecision
> meet the people who need to get their work done yesterday and the people
> who have a lot of experience with large scale deployment, but in a company
> culture that's not representative of the C++ community as a whole, and the
> end result is supposed to be the best of these worlds.
> 
> But it isn't.

The idea that disappointment with the standards process more than one 
cause and different libraries disappoint for different reasons goes a 
long way to explain why we're having so much trouble suggesting 
improvements to the process.

Robert Ramey